BONES-SEED Dataset License Agreement

THIS DATASET LICENSE AGREEMENT (THE "AGREEMENT") IS A LEGAL AGREEMENT BETWEEN AIDVISORY SP. Z O.O. ("LICENSOR") AND THE ENTITY YOU REPRESENT ("LICENSEE").

BY CLICKING "AGREE AND SEND REQUEST TO ACCESS REPO" ON THE HUGGING FACE PLATFORM, OR BY DOWNLOADING OR USING THE DATASET VIA AN API TOKEN OR SCRIPT, LICENSEE AGREES TO BE BOUND BY THESE TERMS. IF YOU DO NOT AGREE, DO NOT CLICK THE BUTTON AND DO NOT ACCESS THE DATASET.

ELIGIBILITY NOTICE. If you do not qualify as (i) an Academic User or (ii) a Qualifying Startup, then you are NOT granted any rights under this Agreement and you must not access, download, or use the Dataset. If you wish to use the Dataset, you must obtain a separate commercial license by contacting: licensing@bones.studio

1. Definitions

  • Academic User: A Licensee who is a non-profit academic institution (such as a university or research institute), using the Dataset strictly for non-commercial scientific research that is intended to be and is subsequently made publicly available (e.g., via Open Access publication), excluding researchers employed by or affiliated with commercial entities other than Qualifying Startups, even if they also hold an academic affiliation.
  • Affiliate: Any entity that directly or indirectly controls, is controlled by, or is under common control with the Licensee. For purposes of this definition, "control" means the power to direct the management and policies of an entity, whether through ownership of voting securities, by contract, or otherwise.
  • Change of Control: Any merger, acquisition, IPO, or sale of all or substantially all of Licensee’s assets, or any transaction resulting in a transfer of more than 50% of the voting power of Licensee.
  • Competing Dataset: Any dataset substantially similar to the Dataset in scope or intended use that is marketed, licensed, or made available to third parties.
  • Dataset: The BONES-SEED proprietary motion capture data provided by Licensor, including any associated documentation or metadata including but not limited to the Dataset specifications described in Schedule 1 hereto. The Dataset consists of elements protected by copyright and related rights, database rights, and/or other proprietary rights, including but not limited to original taxonomy and classification, structure, database schema, and descriptions. The Dataset constitutes confidential information of the Licensor.
  • Qualifying Startup: A Licensee who is a commercial entity which, at the time of accessing the Dataset has an annual gross revenue of less than $1,000,000 USD.
  • Results: Any publicly or privately distributed output created by Licensee that is based upon or incorporates the Dataset, such as trained machine learning models, rendered animations, statistical findings, publications. Results include any derivative works based on the Dataset.

2. Grant of License

Subject to the terms of this Agreement, Licensor grants the Licensee a non-exclusive, non-transferable, revocable license to use the Dataset solely in accordance with the terms and conditions set forth herein and for the sole purpose permitted by Licensee's eligibility classification (Section 3). The license covers the following fields of use: (a) fixation, reproduction, and processing: the right to fix, reproduce, process, transform, and computationally analyze the Dataset, in whole or in part, by any technique—including digital techniques—in the memory of a computer, server, or cloud storage, solely for the purpose of generating permitted Results; such use, subject to the restrictions in Section 6, includes training, fine-tuning, and evaluation of machine learning models.; (b) Public Display: the right to publicly display portions of the Dataset solely as part of the permitted Results, provided the raw data is not accessible.

3. Eligibility and Scope of Use

If you are accepting these terms on behalf of an entity (e.g., a university, research institute, or startup company), You represent and warrant that you have the legal authority to bind such entity to this Agreement. "Licensee" shall refer to that entity.

The Licensee must represent and warrant that they meet the criteria of one of the following categories:

  • Academic User: Use is strictly limited to non-commercial scientific research, evaluation, and educational purposes.
  • Qualifying Startup: Use is permitted solely for Licensee’s internal research, development, and commercialization activities, provided that Licensee continues to meet the eligibility criteria of a Qualifying Startup as defined in Section 1.

4. Change of Status

A Qualifying Startup Licensee must notify Licensor via email at [NOTIFICATION EMAIL] within thirty (30) days if: (a) Licensee’s annual gross revenue exceeds $1,000,000 USD; or (b) Licensee undergoes a Change of Control.

The license granted herein shall automatically terminate thirty (30) days after the date of such event. To continue using the Dataset and/or Results, Licensee must enter into Licensor’s standard commercial license agreement and pay the applicable fees in accordance with the current price list published at www.bones.studio/licensing . Continued use without such conversion constitutes a material breach and intellectual property rights infringement.

Licensor or its designated agent may, upon reasonable notice and during normal business hours, audit Licensee’s records and use of the Dataset to verify compliance with the eligibility criteria (Section 3) and the terms of this Agreement. If an audit reveals that Licensee did not meet the Qualifying Startup criteria at the time of use, Licensee shall reimburse Licensor for the cost of the audit and [immediately pay the standard commercial license fees applicable to such use, plus the maximum statutory interest for delay in commercial transactions calculated from the date such fees became due].

Nothing in this Section limits Licensor’s right to pursue additional remedies, including but not limited to for intellectual property rights infringement.

5. Intellectual Property and Ownership

5.1. Licensor Ownership

Licensor retains all right, title, and interest in and to the raw Dataset, including any and all copies, modifications, and derivative works of the raw Dataset itself (excluding the Licensee's Results defined below). This includes all associated intellectual property rights. Licensee acknowledges that the Dataset is a protected database within the meaning of applicable database protection laws. Licensor is the sole producer of the database and retains all sui generis database rights, copyright, and other proprietary rights in and to the raw Database. All rights not expressly granted are reserved.

5.2. Licensee Ownership

Subject to Licensee’s compliance with this Agreement, Licensee acquires all proprietary copyrights to any Results created by Licensee. To the extent that such Results constitute derivative works of the Dataset, Licensor hereby grants Licensee permission to exercise derivative copyright in such Results. This permission is valid solely provided that the raw Dataset itself cannot be extracted, reverse-engineered, or re-obtained from such Results. For example, a trained machine learning model is owned by the Licensee, but the raw motion capture files used to train it remain the property of the Licensor. For clarity, ownership of Results does not grant any right to redistribute the Dataset or its data.

5.3. Feedback

If Licensee provides any suggestions, feedback, or improvements regarding the Dataset, Licensee hereby grants Licensor a perpetual, irrevocable, royalty-free, worldwide license to use, modify, and incorporate them into Licensor’s products and services without restriction or compensation, including the fields of use set out in Section 2.

6. Restrictions on Use

Licensee shall not, and shall not permit any third party to:

  • No Redistribution: Sell, rent, lease, lend, license, sublicense, redistribute, allow the use or otherwise transfer the raw Dataset or any portion thereof to any Affiliate or third party. Any breach of this Agreement by an Affiliate shall be deemed a breach by the Licensee.
  • No Re-identification: Attempt to re-identify any actors, subjects, or individuals within the motion data, even if the data is anonymized.
  • Prohibited Uses: Use the Dataset for any unlawful purpose, to create or distribute deepfakes, or for malicious surveillance, monitoring, or tracking purposes.
  • No Reverse Engineering: Reverse engineer, decompile, or disassemble the Dataset, except as explicitly permitted by applicable statutory law.
  • No High-Risk Use: Use the Dataset or any Results in any hazardous environments requiring fail-safe performance, including but not limited to the operation of weapons systems, critical infrastructure or medical equipment.
  • No Removal of Attribution: Remove, obscure, or fail to provide attribution as required by this Agreement.
  • No Public Mirrors or Forks: Upload the Dataset to a public repository on the Hugging Face platform, GitHub, or any other third-party platform. "Forking" or copying the Dataset to a private repository is permitted solely for internal use, provided that such repository remains private and inaccessible to third parties.
  • No Competing Datasets & Generative Models:Use the Dataset or Results to create, improve, or augment a Competing Dataset or a commercially substitutable synthetic dataset. This includes, without limitation, using the Dataset to train, fine-tune, or condition any generative model whose primary purpose or substantial use is to generate motion capture data, human motion sequences, animation data, or any other output that functions as a commercial substitute for the Dataset or any portion thereof. For the avoidance of doubt, this restriction does not prohibit training machine learning models (such as robotic control policies, vision-language-action models, behavior cloning models, or motion tracking models) that consume the Dataset as training data and produce non-data outputs (such as control signals, predictions, or classifications). For purposes of this clause, an output "functions as a commercial substitute" if it could reasonably be licensed, sold, or distributed to third parties as a replacement for the Dataset or used to reduce or eliminate the need to license the Dataset.
  • No Work-for-Hire: Use the Dataset to perform development or model training services for any third party that does not independently meet the eligibility criteria of this Agreement, where such third party obtains ownership of the Results. For clarity, selling Licensee’s standard off-the-shelf products to third parties is permitted.

Nothing in this Agreement prohibits any act that cannot be prohibited under applicable law.

7. Confidentiality

Licensee agrees to hold the Dataset in strict confidence and shall not disclose, publish, or otherwise make the raw Dataset available to any third party, except to its employees or contractors who have a strict "need to know" for the permitted purpose and who are bound by written confidentiality obligations at least as restrictive as those contained herein.

The obligations in this Section 7 do not apply to information that: (a) becomes publicly known through no fault of Licensee (e.g., if Licensor releases the data publicly); or (b) is required to be disclosed by law or court order.

8. Attribution, Publicity

In any publication, product, academic paper, technical documentation, or public display of Results derived from the Dataset, the Licensee must provide visible, professional credit to the Licensor.For papers and technical reports, include the credit in the Acknowledgements or Data section. For products, demos, and websites, include the credit on an About/Credits page or other prominent location. The specific citation format is: Motion Data by Bones Studio with a link to https://bones.studio/.

For machine learning models and software: include the following attribution in the model card (e.g., HuggingFace README.md), repository README, or equivalent documentation that accompanies the distribution of the model or software: Training data includes Motion Data by Bones Studio with a link to https://bones.studio/. Use of the underlying dataset is subject to the BONES Motion Capture Dataset License Agreement.

Failure to provide attribution is a material breach.

Licensor may identify Licensee as a user of the Dataset and may use Licensee’s name and logo in customer lists and marketing materials, case studies or joint announcements referencing Licensee’s use of the Dataset, provided that Licensor first informs Licensee in writing (email sufficient). Licensor may proceed with such use unless Licensee submits a written objection (email sufficient) within seven (7) days of receiving said information.

9. Warranties and Disclaimers

THE DATASET IS PROVIDED "AS IS" AND "WITH ALL FAULTS." LICENSOR MAKES NO WARRANTIES, EXPRESS OR IMPLIED, REGARDING THE DATASET. TO THE FULLEST EXTENT PERMITTED BY LAW, THE PARTIES HEREBY EXCLUDE LICENSOR’S LIABILITY UNDER STATUTORY WARRANTY FOR DEFECTS. LICENSOR DOES NOT WARRANT THAT THE DATASET WILL BE ERROR-FREE OR THAT THE USE OF THE DATASET WILL BE UNINTERRUPTED.

Licensor is under no obligation to provide updates or upgrades to the Dataset.

10. Limitation of Liability

TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, LICENSOR SHALL NOT BE LIABLE FOR ANY LOST PROFITS, LOSS OF REVENUE, LOSS OF DATA, OR ANY INDIRECT, INCIDENTAL, OR CONSEQUENTIAL DAMAGES. LICENSOR’S TOTAL CUMULATIVE LIABILITY ARISING OUT OF OR RELATED TO THIS AGREEMENT WILL BE LIMITED TO ONE HUNDRED DOLLARS ($100 USD). THE FOREGOING LIMITATIONS SHALL NOT APPLY TO DAMAGES CAUSED BY LICENSOR’S INTENTIONAL MISCONDUCT.

11. Indemnification

Licensee agrees to indemnify, defend, and hold harmless Licensor, its officers, directors, employees, and agents from and against any and all claims, damages, liabilities, costs, and expenses (including reasonable attorneys' fees) arising out of or related to Licensee's use of the Dataset or Licensee's breach of any term of this Agreement.

12. Termination

Licensor may terminate this Agreement immediately upon notice in document form (email sufficient) if Licensee breaches any material term. Upon termination, the license granted herein is automatically revoked. Licensee must immediately: (i) cease all use of the Dataset; (ii) cease all use, reproduction, distribution, and commercialization of any Results generated using the Dataset, and withdraw such Results from any products or services (Licensee acknowledges that the permission to exercise derivative copyright granted in Section 5.2 is automatically revoked upon termination); (iii) permanently delete or destroy all copies of the raw Dataset in its possession or control; and (iv) certify such destruction in document form (e.g., via email) to Licensor within five (5) business days. The provisions regarding Warranties, Intellectual Property, Limitation of Liability, Indemnification, Termination and Governing Law and Venue shall survive termination.

13. Governing Law and Venue

This Agreement shall be governed by and construed in accordance with the laws of Poland. The application of the United Nations Convention on Contracts for the International Sale of Goods (CISG) is hereby expressly excluded. Any dispute arising out of or related to this Agreement shall be submitted to the exclusive jurisdiction of the courts competent for the seat of the Licensor.

14. Export Control

Licensee represents and warrants that it is not located in, under the control of, or a national or resident of any country or person subject to EU, UK, or U.S. sanctions or export restrictions. Licensee agrees to comply with all applicable export control and sanctions laws and regulations. Licensor may suspend performance immediately if it reasonably determines Licensee’s use would cause a violation.

15. Miscellaneous

15.1. Entire Agreement

This Agreement constitutes the entire understanding between the parties concerning the subject matter herein and supersedes all prior agreements and understandings, whether written or oral.

15.2. No Waiver

The failure of either party to enforce any rights granted hereunder or to take action against the other party in the event of any breach shall not be deemed a waiver by that party as to subsequent enforcement of rights or subsequent actions in the event of future breaches.

15.3. No Assignment

Licensee may not assign or transfer this Agreement, by operation of law or otherwise, without Licensor’s prior written consent. Any attempt to do so is void.

15.4. Severability

If any provision of this Agreement is held to be unenforceable, such provision shall be modified to the minimum extent necessary to make it enforceable, and the remaining provisions will remain in full force and effect.

15.5. Force Majeure

Neither party is liable for any delay or failure to perform due to causes beyond its reasonable control, including acts of God, war, terrorism, civil unrest, labor disputes, embargoes, power failures, internet outages, or governmental actions; provided that the affected party uses commercially reasonable efforts to mitigate and resume performance.

15.6. Modifications

Licensor reserves the right to modify this Agreement. For material changes (e.g., changes to eligibility criteria or restrictions), Licensor will provide at least fourteen (14) days' prior notice via email to the address associated with Licensee’s account or by a prominent notice on the Dataset repository page. Licensee’s continued use of the Dataset after such notice period constitutes acceptance of the revised terms. If Licensee does not agree to the new terms, Licensee must immediately cease using the Dataset and delete all copies as set forth in Section 12.

15.7. Licensee information

Licensee acknowledges that Licensor has access to the user profile information (including email address and username) associated with the Hugging Face account used to access the Dataset. Licensee consents to Licensor using this information solely for the purpose of verifying compliance with this Agreement and contacting Licensee regarding their eligibility status.

Schedule 1

DATASET SPECIFICATIONS

The motion sequences contained in the Dataset are original creative works produced by Licensor. These sequences were specifically scripted, directed, choreographed, and recorded by Licensor in a professional studio environment. They constitute distinct artistic performances and are not mere recordings of random or public domain physical actions.

Motion data assets

The Dataset consists of a total of 142 000 distinct motion capture sequences, provided in the following formats:

  1. SOMA Skeleton (Original Capture Proportions):
    1. Files: .bvh and .npz .
    2. Description: Motion data retargeted to the SOMA skeleton, preserving the limb lengths and proportions of the original performers.
  2. SOMA Skeleton (Uniform Proportions):
    1. Files: .bvh and .npz .
    2. Description: Motion data retargeted to a uniform SOMA skeleton.
  3. Unitree G1 Robot Format:
    1. Files: .csv
    2. Description: Motion trajectories optimized for Unitree G1 humanoid robot, compatible with MuJoCo.

Metadata & annotations

The Dataset is accompanied by metadata files organized according to Licensor’s proprietary taxonomy and structure, as follows:

  1. Semantic metadata with labeling (.csv): a structured dataset containing semantic descriptions, categories and performance info.
  2. Multi-variant language descriptions: each motion is described in up to 6 different ways (natural language descriptions 1-4, technical and short description). These intentionally varied paraphrases of the same physical action constitute original creative work.
  3. Hierarchical categorization system: the Dataset uses a multi-level classification system (categories, styles, motion classifications) developed by Licensor to systematically organize complex human movements.
  4. Skeletal data: The index includes detailed anatomical measurements of source performers (including specific bone dimensions, e.g., collarbone_span, wrist_span), enabling precise correlation between a performer's body morphology and the resulting kinematic data.
  5. Proprietary database schema: the selection of attributes, the hierarchy of descriptive tags, and the compilation of language variants constitute Licensor's proprietary database schema.
  6. Temporal timeline labeling (.json): files containing temporal semantic segmentation data, which divide continuous motion sequences into distinct, meaningful actions.
  7. Segmentation methodology: the Dataset utilizes a structured approach to event segmentation, where motion is partitioned into logical chronological phases (e.g., approach, interaction, retreat) defined by precise start and end timestamps.
  8. Narrative description: each segment is paired with a natural language description that interprets the specific physical action, context, and causal factors (e.g., "leans back due to the weight"). The specific composition of these descriptive narratives and their precise synchronization with motion events constitutes a proprietary creative work owned by Licensor.